Quantitative structure-activity relationship study on the anti-HIV-1 activity of novel 6-naphthylthio HEPT analogs

The quantitative structure-activity relationship of the novel 6-naphthylthio 1-[(2-hydroxyethoxy)methyl]-6-(phenylthio) thymine derivatives for prediction of anti-human immunodeficiency virus type 1 activity was studied. The suitable set of the molecular descriptors was calculated and the important descriptors using the variable selections of the stepwise multiple linear regression and the genetic algorithm were selected. A comparison between the attained results indicated the superiority of the genetic algorithm over the stepwise multiple regression method in the feature-selection. The predictive quality of the quantitative structure-activity relationship models was tested for an external set of eight compounds, randomly chosen out of 39 compounds. The genetic algorithm-multiple linear regression model with six selected descriptors was obtained. This model, demonstrating high statistical qualities (R(2)(train) = 0.925, Q(2) = 0.872, SE (%) = 1.23, F = 49.338, R(2)(pred) = 0.944), could predict the anti-human immunodeficiency virus type 1 activity of the molecules with a prediction error percentage lower than 10%. The results suggest that electronegativity, the masses, and the atomic van der Waals volumes are the main independent factors contributing to the anti-human immunodeficiency virus type 1 activity of the studied compounds.